# импортируем необходимые объекты и модули
from tensorflow.keras.applications.resnet import ResNet50
from tensorflow.keras.preprocessing.image import ImageDataGenerator
from tensorflow.keras.layers import GlobalAveragePooling2D, Dense
from tensorflow.keras.models import Sequential
from tensorflow.keras.optimizers import Adam
import numpy as np
import pandas as pd
import tensorflow as tf
# создание загрузчика для обучающей выборки с отражением фотографий по горизонтали
def load_train(path):
labels = pd.read_csv(path + 'labels.csv')
datagen = ImageDataGenerator(validation_split=0.25, horizontal_flip=True, rescale=1./255)
train_gen_flow = datagen.flow_from_dataframe(
dataframe=labels,
directory=path + 'final_files/',
x_col='file_name',
y_col='real_age',
target_size=(224, 224),
batch_size=16,
class_mode='raw',
subset='training',
seed=12345)
return train_gen_flow
# создание загрузчика для тестовой выборки
def load_test(path):
labels = pd.read_csv(path + 'labels.csv')
datagen = ImageDataGenerator(validation_split=0.25, rescale=1./255)
test_gen_flow = datagen.flow_from_dataframe(
dataframe=labels,
directory=path + 'final_files/',
x_col='file_name',
y_col='real_age',
target_size=(224, 224),
batch_size=16,
class_mode='raw',
subset='validation',
seed=12345)
return test_gen_flow
# создание модели
def create_model(input_shape):
backbone = ResNet50(input_shape=input_shape,
weights='imagenet',
include_top=False)
model = Sequential()
model.add(backbone)
model.add(GlobalAveragePooling2D())
model.add(Dense(1, activation='relu'))
optimizer = Adam(lr=0.0001)
model.compile(optimizer=optimizer, loss='mean_squared_error',
metrics=['mae'])
return model
# параметры обучения
def train_model(model, train_data, test_data, batch_size=None, epochs=10,
steps_per_epoch=None, validation_steps=None):
if steps_per_epoch is None:
steps_per_epoch = len(train_data)
if validation_steps is None:
validation_steps = len(test_data)
model.fit(train_data,
validation_data=test_data,
batch_size=batch_size,
epochs=epochs,
steps_per_epoch=steps_per_epoch,
validation_steps=validation_steps,
verbose=2)
return model
